Crowd Dynamics: Social Force Simulator

Crowd Dynamics

Crowd size 160

Panic level 0.3

Exit width 2.2 m

Avg flow speed

–

Throughput

–

Exit congestion

–

Evacuated

–

How it works

Every agent is pulled toward a desired velocity — its personal walking speed, aimed at the doorway — and pushed away from every neighbour close enough to feel crowded, plus the walls. This is the social force model (Helbing & Molnár): a purely local rule, repeated across hundreds of agents, that produces the lane formation, arching and clogging seen in real crowds without anyone planning it centrally.

F_i = (v0_i − v_i)/τ + Σⱼ A·exp((rᵢⱼ−dᵢⱼ)/B)·n̂ᵢⱼ + F_wall
v0_i = desiredSpeed · panic · direction(exit)

Crowd size — how many agents share the room; more agents mean more overlapping repulsion forces and a higher chance of jamming near the door.
Panic level — raises each agent's desired speed and shrinks its comfortable personal-space radius. Past a threshold this triggers the "faster is slower" effect: everyone pushes harder, an arch of bodies locks at the doorway, and total throughput actually drops.
Exit width — a wider doorway lets more agents pass side-by-side before their repulsion fields overlap and jam.
Herding — toggles a small alignment force that nudges each agent toward the average heading of its nearby neighbours, mimicking the social contagion that makes a crowd's direction changes ripple outward from a few leaders.

Real-world relevance: this same model underlies stadium and transit-station evacuation planning — architects widen doorways and add central pillars near exits specifically to break up the arching that panic accelerates rather than relieves.

Flowing freely Congested / pushing Leader (herding)

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.